[yt-dlp.git] / yt_dlp / extractor / mediastream.py

import re

from .common import InfoExtractor
from ..utils import (
    clean_html,
    remove_end,
    traverse_obj,
    urljoin,
)


class MediaStreamBaseIE(InfoExtractor):
    _EMBED_BASE_URL = 'https://mdstrm.com/embed'
    _BASE_URL_RE = r'https?://mdstrm\.com/(?:embed|live-stream)'

    def _extract_mediastream_urls(self, webpage):
        yield from traverse_obj(list(self._yield_json_ld(webpage, None, fatal=False)), (
            lambda _, v: v['@type'] == 'VideoObject', ('embedUrl', 'contentUrl'),
            {lambda x: x if re.match(rf'{self._BASE_URL_RE}/\w+', x) else None}))

        for mobj in re.finditer(r'<script[^>]+>[^>]*playerMdStream\.mdstreamVideo\(\s*[\'"](?P<video_id>\w+)', webpage):
            yield f'{self._EMBED_BASE_URL}/{mobj.group("video_id")}'

        yield from re.findall(
            rf'<iframe[^>]+\bsrc="({self._BASE_URL_RE}/\w+)', webpage)

        for mobj in re.finditer(
            r'''(?x)
                <(?:div|ps-mediastream)[^>]+
                (class="[^"]*MediaStreamVideoPlayer)[^"]*"[^>]+
                data-video-id="(?P<video_id>\w+)"
                (?:\s*data-video-type="(?P<video_type>[^"]+))?
                (?:[^>]*>\s*<div[^>]+\1[^"]*"[^>]+data-mediastream=["\'][^>]+
                    https://mdstrm\.com/(?P<live>live-stream))?
                ''', webpage):

            video_type = 'live-stream' if mobj.group('video_type') == 'live' or mobj.group('live') else 'embed'
            yield f'https://mdstrm.com/{video_type}/{mobj.group("video_id")}'


class MediaStreamIE(MediaStreamBaseIE):
    _VALID_URL = MediaStreamBaseIE._BASE_URL_RE + r'/(?P<id>\w+)'

    _TESTS = [{
        'url': 'https://mdstrm.com/embed/6318e3f1d1d316083ae48831',
        'md5': '97b4f2634b8e8612cc574dfcd504df05',
        'info_dict': {
            'id': '6318e3f1d1d316083ae48831',
            'title': 'Video: Así fue el despido de Thomas Tuchel del Chelsea',
            'description': 'md5:358ce1e1396010d50a1ece1be3633c95',
            'thumbnail': r're:^https?://[^?#]+6318e3f1d1d316083ae48831',
            'ext': 'mp4',
        },
        'params': {'skip_download': 'm3u8'},
    }]

    _WEBPAGE_TESTS = [{
        'url': 'https://www.multimedios.com/video/costa-rica-tv-en-vivo/v2616',
        'info_dict': {
            'id': '5a7b1e63a8da282c34d65445',
            'title': 're:mmtv-costarica',
            'description': 'mmtv-costarica',
            'thumbnail': 're:^https?://[^?#]+5a7b1e63a8da282c34d65445',
            'ext': 'mp4',
            'live_status': 'is_live',
        },
        'params': {'skip_download': 'Livestream'},
    }, {
        'url': 'https://www.multimedios.com/television/clases-de-llaves-y-castigos-quien-sabe-mas',
        'md5': 'de31f0b1ecc321fb35bf22d58734ea40',
        'info_dict': {
            'id': '63731bab8ec9b308a2c9ed28',
            'title': 'Clases de llaves y castigos ¿Quién sabe más?',
            'description': 'md5:1b49aa1ee5a4b32fbd66104b2d629e9d',
            'thumbnail': 're:^https?://[^?#]+63731bab8ec9b308a2c9ed28',
            'ext': 'mp4',
        },
        'params': {'skip_download': 'm3u8'},
    }, {
        'url': 'https://www.americatv.com.pe/videos/esto-es-guerra/facundo-gonzalez-sufrio-fuerte-golpe-durante-competencia-frente-hugo-garcia-eeg-noticia-139120',
        'info_dict': {
            'id': '63756df1c638b008a5659dec',
            'title': 'Facundo González sufrió fuerte golpe durante competencia frente a Hugo García en EEG',
            'description': 'md5:9490c034264afd756eef7b2c3adee69e',
            'thumbnail': 're:^https?://[^?#]+63756df1c638b008a5659dec',
            'ext': 'mp4',
        },
        'params': {'skip_download': 'm3u8'},
    }, {
        'url': 'https://www.americatv.com.pe/videos/al-fondo-hay-sitio/nuevas-lomas-town-bernardo-mata-se-enfrento-sujeto-luchar-amor-macarena-noticia-139083',
        'info_dict': {
            'id': '637307669609130f74cd3a6e',
            'title': 'Las Nuevas Lomas Town: Bernardo De La Mata se enfrentó a sujeto para luchar por el amor de Macarena',
            'description': 'md5:60d71772f1e1496923539ae58aa17124',
            'thumbnail': 're:^https?://[^?#]+637307669609130f74cd3a6e',
            'ext': 'mp4',
        },
        'params': {'skip_download': 'm3u8'},
    }]

    def _extract_from_webpage(self, url, webpage):
        for embed_url in self._extract_mediastream_urls(webpage):
            yield self.url_result(embed_url, MediaStreamIE, None)

    def _real_extract(self, url):
        video_id = self._match_id(url)
        webpage = self._download_webpage(url, video_id)

        for message in [
            'Debido a tu ubicación no puedes ver el contenido',
            'You are not allowed to watch this video: Geo Fencing Restriction'
        ]:
            if message in webpage:
                self.raise_geo_restricted()

        player_config = self._search_json(r'window\.MDSTRM\.OPTIONS\s*=', webpage, 'metadata', video_id)

        formats, subtitles = [], {}
        for video_format in player_config['src']:
            if video_format == 'hls':
                fmts, subs = self._extract_m3u8_formats_and_subtitles(player_config['src'][video_format], video_id)
                formats.extend(fmts)
                self._merge_subtitles(subs, target=subtitles)
            elif video_format == 'mpd':
                fmts, subs = self._extract_mpd_formats_and_subtitles(player_config['src'][video_format], video_id)
                formats.extend(fmts)
                self._merge_subtitles(subs, target=subtitles)
            else:
                formats.append({
                    'url': player_config['src'][video_format],
                })

        return {
            'id': video_id,
            'title': self._og_search_title(webpage) or player_config.get('title'),
            'description': self._og_search_description(webpage),
            'formats': formats,
            'subtitles': subtitles,
            'is_live': player_config.get('type') == 'live',
            'thumbnail': self._og_search_thumbnail(webpage),
        }


class WinSportsVideoIE(MediaStreamBaseIE):
    _VALID_URL = r'https?://www\.winsports\.co/videos/(?P<id>[\w-]+)'

    _TESTS = [{
        'url': 'https://www.winsports.co/videos/siempre-castellanos-gran-atajada-del-portero-cardenal-para-evitar-la-caida-de-su-arco-60536',
        'info_dict': {
            'id': '62dc8357162c4b0821fcfb3c',
            'display_id': 'siempre-castellanos-gran-atajada-del-portero-cardenal-para-evitar-la-caida-de-su-arco-60536',
            'title': '¡Siempre Castellanos! Gran atajada del portero \'cardenal\' para evitar la caída de su arco',
            'description': 'md5:eb811b2b2882bdc59431732c06b905f2',
            'thumbnail': r're:^https?://[^?#]+62dc8357162c4b0821fcfb3c',
            'ext': 'mp4',
        },
        'params': {'skip_download': 'm3u8'},
    }, {
        'url': 'https://www.winsports.co/videos/observa-aqui-los-goles-del-empate-entre-tolima-y-nacional-60548',
        'info_dict': {
            'id': '62dcb875ef12a5526790b552',
            'display_id': 'observa-aqui-los-goles-del-empate-entre-tolima-y-nacional-60548',
            'title': 'Observa aquí los goles del empate entre Tolima y Nacional',
            'description': 'md5:b19402ba6e46558b93fd24b873eea9c9',
            'thumbnail': r're:^https?://[^?#]+62dcb875ef12a5526790b552',
            'ext': 'mp4',
        },
        'params': {'skip_download': 'm3u8'},
    }, {
        'url': 'https://www.winsports.co/videos/equidad-vuelve-defender-su-arco-de-remates-de-junior',
        'info_dict': {
            'id': '63fa7eca72f1741ad3a4d515',
            'display_id': 'equidad-vuelve-defender-su-arco-de-remates-de-junior',
            'title': '⚽ Equidad vuelve a defender su arco de remates de Junior',
            'description': 'Remate de Sierra',
            'thumbnail': r're:^https?://[^?#]+63fa7eca72f1741ad3a4d515',
            'ext': 'mp4',
        },
        'params': {'skip_download': 'm3u8'},
    }, {
        'url': 'https://www.winsports.co/videos/bucaramanga-se-quedo-con-el-grito-de-gol-en-la-garganta',
        'info_dict': {
            'id': '6402adb62bbf3b18d454e1b0',
            'display_id': 'bucaramanga-se-quedo-con-el-grito-de-gol-en-la-garganta',
            'title': '⚽Bucaramanga se quedó con el grito de gol en la garganta',
            'description': 'Gol anulado Bucaramanga',
            'thumbnail': r're:^https?://[^?#]+6402adb62bbf3b18d454e1b0',
            'ext': 'mp4',
        },
        'params': {'skip_download': 'm3u8'},
    }]

    def _real_extract(self, url):
        display_id = self._match_id(url)
        webpage = self._download_webpage(url, display_id)
        data = self._search_json(
            r'<script\s*[^>]+data-drupal-selector="drupal-settings-json">', webpage, 'data', display_id)

        mediastream_url = urljoin(f'{self._EMBED_BASE_URL}/', (
            traverse_obj(data, (
                (('settings', 'mediastream_formatter', ..., 'mediastream_id'), 'url'), {str}), get_all=False)
            or next(self._extract_mediastream_urls(webpage), None)))

        if not mediastream_url:
            self.raise_no_formats('No MediaStream embed found in webpage')

        title = clean_html(remove_end(
            self._search_json_ld(webpage, display_id, expected_type='VideoObject', default={}).get('title')
            or self._og_search_title(webpage), '| Win Sports'))

        return self.url_result(
            mediastream_url, MediaStreamIE, display_id, url_transparent=True, display_id=display_id, video_title=title)
Commit	Line	Data
3d79ebc8 E	1	import re
	2
	3	from .common import InfoExtractor
2d5a8c5d	4	from ..utils import (
03025b6e	5	clean_html,
2d5a8c5d	6	remove_end,
2d5a8c5d	7	traverse_obj,
	8	urljoin,
	9	)
3d79ebc8 E	10
3d79ebc8 E	11
03025b6e	12	class MediaStreamBaseIE(InfoExtractor):
	13	_EMBED_BASE_URL = 'https://mdstrm.com/embed'
	14	_BASE_URL_RE = r'https?://mdstrm\.com/(?:embed\|live-stream)'
	15
	16	def _extract_mediastream_urls(self, webpage):
635ae31f	17	yield from traverse_obj(list(self._yield_json_ld(webpage, None, fatal=False)), (
03025b6e	18	lambda _, v: v['@type'] == 'VideoObject', ('embedUrl', 'contentUrl'),
	19	{lambda x: x if re.match(rf'{self._BASE_URL_RE}/\w+', x) else None}))
	20
	21	for mobj in re.finditer(r'<script[^>]+>[^>]playerMdStream\.mdstreamVideo\(\s[\'"](?P<video_id>\w+)', webpage):
	22	yield f'{self._EMBED_BASE_URL}/{mobj.group("video_id")}'
	23
	24	yield from re.findall(
	25	rf'<iframe[^>]+\bsrc="({self._BASE_URL_RE}/\w+)', webpage)
	26
	27	for mobj in re.finditer(
	28	r'''(?x)
	29	<(?:div\|ps-mediastream)[^>]+
	30	(class="[^"]MediaStreamVideoPlayer)[^"]"[^>]+
	31	data-video-id="(?P<video_id>\w+)"
	32	(?:\s*data-video-type="(?P<video_type>[^"]+))?
	33	(?:[^>]>\s<div[^>]+\1[^"]*"[^>]+data-mediastream=["\'][^>]+
	34	https://mdstrm\.com/(?P<live>live-stream))?
	35	''', webpage):
	36
	37	video_type = 'live-stream' if mobj.group('video_type') == 'live' or mobj.group('live') else 'embed'
	38	yield f'https://mdstrm.com/{video_type}/{mobj.group("video_id")}'
	39
	40
	41	class MediaStreamIE(MediaStreamBaseIE):
	42	_VALID_URL = MediaStreamBaseIE._BASE_URL_RE + r'/(?P<id>\w+)'
3d79ebc8 E	43
	44	_TESTS = [{
	45	'url': 'https://mdstrm.com/embed/6318e3f1d1d316083ae48831',
	46	'md5': '97b4f2634b8e8612cc574dfcd504df05',
	47	'info_dict': {
	48	'id': '6318e3f1d1d316083ae48831',
	49	'title': 'Video: Así fue el despido de Thomas Tuchel del Chelsea',
	50	'description': 'md5:358ce1e1396010d50a1ece1be3633c95',
	51	'thumbnail': r're:^https?://[^?#]+6318e3f1d1d316083ae48831',
	52	'ext': 'mp4',
	53	},
03025b6e	54	'params': {'skip_download': 'm3u8'},
3d79ebc8 E	55	}]
	56
	57	_WEBPAGE_TESTS = [{
	58	'url': 'https://www.multimedios.com/video/costa-rica-tv-en-vivo/v2616',
	59	'info_dict': {
	60	'id': '5a7b1e63a8da282c34d65445',
	61	'title': 're:mmtv-costarica',
	62	'description': 'mmtv-costarica',
	63	'thumbnail': 're:^https?://[^?#]+5a7b1e63a8da282c34d65445',
	64	'ext': 'mp4',
	65	'live_status': 'is_live',
	66	},
03025b6e	67	'params': {'skip_download': 'Livestream'},
3d79ebc8 E	68	}, {
	69	'url': 'https://www.multimedios.com/television/clases-de-llaves-y-castigos-quien-sabe-mas',
	70	'md5': 'de31f0b1ecc321fb35bf22d58734ea40',
	71	'info_dict': {
	72	'id': '63731bab8ec9b308a2c9ed28',
	73	'title': 'Clases de llaves y castigos ¿Quién sabe más?',
	74	'description': 'md5:1b49aa1ee5a4b32fbd66104b2d629e9d',
	75	'thumbnail': 're:^https?://[^?#]+63731bab8ec9b308a2c9ed28',
	76	'ext': 'mp4',
	77	},
03025b6e	78	'params': {'skip_download': 'm3u8'},
3d79ebc8 E	79	}, {
	80	'url': 'https://www.americatv.com.pe/videos/esto-es-guerra/facundo-gonzalez-sufrio-fuerte-golpe-durante-competencia-frente-hugo-garcia-eeg-noticia-139120',
	81	'info_dict': {
	82	'id': '63756df1c638b008a5659dec',
	83	'title': 'Facundo González sufrió fuerte golpe durante competencia frente a Hugo García en EEG',
	84	'description': 'md5:9490c034264afd756eef7b2c3adee69e',
	85	'thumbnail': 're:^https?://[^?#]+63756df1c638b008a5659dec',
	86	'ext': 'mp4',
	87	},
03025b6e	88	'params': {'skip_download': 'm3u8'},
3d79ebc8 E	89	}, {
	90	'url': 'https://www.americatv.com.pe/videos/al-fondo-hay-sitio/nuevas-lomas-town-bernardo-mata-se-enfrento-sujeto-luchar-amor-macarena-noticia-139083',
	91	'info_dict': {
	92	'id': '637307669609130f74cd3a6e',
	93	'title': 'Las Nuevas Lomas Town: Bernardo De La Mata se enfrentó a sujeto para luchar por el amor de Macarena',
	94	'description': 'md5:60d71772f1e1496923539ae58aa17124',
	95	'thumbnail': 're:^https?://[^?#]+637307669609130f74cd3a6e',
	96	'ext': 'mp4',
	97	},
03025b6e	98	'params': {'skip_download': 'm3u8'},
3d79ebc8 E	99	}]
3d79ebc8 E	100
03025b6e	101	def _extract_from_webpage(self, url, webpage):
	102	for embed_url in self._extract_mediastream_urls(webpage):
	103	yield self.url_result(embed_url, MediaStreamIE, None)
3d79ebc8 E	104
	105	def _real_extract(self, url):
	106	video_id = self._match_id(url)
	107	webpage = self._download_webpage(url, video_id)
	108
b3febedb E	109	for message in [
	110	'Debido a tu ubicación no puedes ver el contenido',
	111	'You are not allowed to watch this video: Geo Fencing Restriction'
	112	]:
	113	if message in webpage:
	114	self.raise_geo_restricted()
3d79ebc8	115
03025b6e	116	player_config = self._search_json(r'window\.MDSTRM\.OPTIONS\s*=', webpage, 'metadata', video_id)
3d79ebc8 E	117
	118	formats, subtitles = [], {}
	119	for video_format in player_config['src']:
	120	if video_format == 'hls':
	121	fmts, subs = self._extract_m3u8_formats_and_subtitles(player_config['src'][video_format], video_id)
	122	formats.extend(fmts)
	123	self._merge_subtitles(subs, target=subtitles)
	124	elif video_format == 'mpd':
	125	fmts, subs = self._extract_mpd_formats_and_subtitles(player_config['src'][video_format], video_id)
	126	formats.extend(fmts)
	127	self._merge_subtitles(subs, target=subtitles)
	128	else:
	129	formats.append({
	130	'url': player_config['src'][video_format],
	131	})
	132
	133	return {
	134	'id': video_id,
	135	'title': self._og_search_title(webpage) or player_config.get('title'),
	136	'description': self._og_search_description(webpage),
	137	'formats': formats,
	138	'subtitles': subtitles,
	139	'is_live': player_config.get('type') == 'live',
	140	'thumbnail': self._og_search_thumbnail(webpage),
	141	}
	142
	143
03025b6e	144	class WinSportsVideoIE(MediaStreamBaseIE):
2d5a8c5d	145	_VALID_URL = r'https?://www\.winsports\.co/videos/(?P<id>[\w-]+)'
3d79ebc8 E	146
	147	_TESTS = [{
	148	'url': 'https://www.winsports.co/videos/siempre-castellanos-gran-atajada-del-portero-cardenal-para-evitar-la-caida-de-su-arco-60536',
	149	'info_dict': {
	150	'id': '62dc8357162c4b0821fcfb3c',
2d5a8c5d	151	'display_id': 'siempre-castellanos-gran-atajada-del-portero-cardenal-para-evitar-la-caida-de-su-arco-60536',
3d79ebc8 E	152	'title': '¡Siempre Castellanos! Gran atajada del portero \'cardenal\' para evitar la caída de su arco',
	153	'description': 'md5:eb811b2b2882bdc59431732c06b905f2',
	154	'thumbnail': r're:^https?://[^?#]+62dc8357162c4b0821fcfb3c',
	155	'ext': 'mp4',
	156	},
2d5a8c5d	157	'params': {'skip_download': 'm3u8'},
3d79ebc8 E	158	}, {
	159	'url': 'https://www.winsports.co/videos/observa-aqui-los-goles-del-empate-entre-tolima-y-nacional-60548',
	160	'info_dict': {
	161	'id': '62dcb875ef12a5526790b552',
2d5a8c5d	162	'display_id': 'observa-aqui-los-goles-del-empate-entre-tolima-y-nacional-60548',
3d79ebc8 E	163	'title': 'Observa aquí los goles del empate entre Tolima y Nacional',
	164	'description': 'md5:b19402ba6e46558b93fd24b873eea9c9',
	165	'thumbnail': r're:^https?://[^?#]+62dcb875ef12a5526790b552',
	166	'ext': 'mp4',
	167	},
2d5a8c5d	168	'params': {'skip_download': 'm3u8'},
	169	}, {
	170	'url': 'https://www.winsports.co/videos/equidad-vuelve-defender-su-arco-de-remates-de-junior',
	171	'info_dict': {
	172	'id': '63fa7eca72f1741ad3a4d515',
	173	'display_id': 'equidad-vuelve-defender-su-arco-de-remates-de-junior',
	174	'title': '⚽ Equidad vuelve a defender su arco de remates de Junior',
	175	'description': 'Remate de Sierra',
	176	'thumbnail': r're:^https?://[^?#]+63fa7eca72f1741ad3a4d515',
	177	'ext': 'mp4',
	178	},
	179	'params': {'skip_download': 'm3u8'},
03025b6e	180	}, {
	181	'url': 'https://www.winsports.co/videos/bucaramanga-se-quedo-con-el-grito-de-gol-en-la-garganta',
	182	'info_dict': {
	183	'id': '6402adb62bbf3b18d454e1b0',
	184	'display_id': 'bucaramanga-se-quedo-con-el-grito-de-gol-en-la-garganta',
	185	'title': '⚽Bucaramanga se quedó con el grito de gol en la garganta',
	186	'description': 'Gol anulado Bucaramanga',
	187	'thumbnail': r're:^https?://[^?#]+6402adb62bbf3b18d454e1b0',
	188	'ext': 'mp4',
	189	},
	190	'params': {'skip_download': 'm3u8'},
3d79ebc8 E	191	}]
	192
	193	def _real_extract(self, url):
2d5a8c5d	194	display_id = self._match_id(url)
3d79ebc8	195	webpage = self._download_webpage(url, display_id)
03025b6e	196	data = self._search_json(
	197	r'<script\s*[^>]+data-drupal-selector="drupal-settings-json">', webpage, 'data', display_id)
	198
	199	mediastream_url = urljoin(f'{self._EMBED_BASE_URL}/', (
	200	traverse_obj(data, (
	201	(('settings', 'mediastream_formatter', ..., 'mediastream_id'), 'url'), {str}), get_all=False)
	202	or next(self._extract_mediastream_urls(webpage), None)))
	203
	204	if not mediastream_url:
2d5a8c5d	205	self.raise_no_formats('No MediaStream embed found in webpage')
3d79ebc8	206
03025b6e	207	title = clean_html(remove_end(
	208	self._search_json_ld(webpage, display_id, expected_type='VideoObject', default={}).get('title')
	209	or self._og_search_title(webpage), '\| Win Sports'))
	210
3d79ebc8	211	return self.url_result(
03025b6e	212	mediastream_url, MediaStreamIE, display_id, url_transparent=True, display_id=display_id, video_title=title)